Content feature selection-based social network competition information blocking method

Abstract

The invention provides a social network competition information blocking method based on content feature selection, and the method comprises the steps: S1, constructing a content-aware multi-competition information propagation model, and enabling a social network to be modeled into a graph structure through the model; s2, on the basis of a content awareness multi-competition information propagation model, calculating the blocking probability that a target node in the network is successfully activated by the limited information but not activated by the competition information; s3, iteratively selecting content features from the candidate content feature set by taking maximization of the blocking probability sum of the nodes of the whole network as a target, and adding the content features into a second content feature set; and S4, generating and deploying restriction information based on the second content feature set to inhibit propagation of competition information. According to the method, under the reality constraint of the fixed initial propagation source, the content features of the restriction information are efficiently selected to maximize the blocking effect of the restriction information on the competition information propagation range, and the problem that the calculation complexity of a traditional algorithm is too high is solved.

Description

Technical Field

[0001] This application relates to the field of social network information dissemination technology, specifically a method for blocking competitive information on social networks based on content feature selection. Background Technology

[0002] As a core channel for modern information dissemination, social networks directly impact business competition, public opinion guidance, and public safety through the breadth and speed of information spread. In many real-world scenarios, multiple pieces of information simultaneously circulate and compete on social networks. Examples include official clarifications vying for public trust against online rumors, or different brand advertisements competing for user attention. Effectively controlling the dissemination of specific information (especially competitive or negative information) is of significant practical importance for businesses and government agencies.

[0003] Traditional research on information dissemination control primarily focuses on "maximizing influence" and its derivative problems, namely, how to maximize the final dissemination range of target information by activating key "seed" nodes in the network within a limited budget (such as selecting a specific number of initial dissemination users). These methods are typically based on classic dissemination models, such as the independent cascade model or the linear threshold model, and employ strategies such as greedy algorithms for node selection. However, this "node selection" paradigm has significant limitations in practical applications: in many pre-defined scenarios (such as information released by an official corporate account or an already launched advertising campaign), the initial source of information dissemination (the set of seed nodes) is fixed and cannot be changed, thus making optimizing node selection impractical.

[0004] To address the aforementioned limitations, researchers have recently proposed a new approach: "maximizing influence based on feature selection." The core of this method lies in optimizing the content features of the information itself (e.g., choosing more attractive hashtags, keywords, or sentiment tendencies) to enhance its dissemination potential when the seed node set for information propagation is fixed. This is more realistic because information publishers can usually control content attributes but have difficulty manipulating user (node) forwarding behavior. However, existing research on content feature selection still has significant shortcomings: First, most current work focuses on increasing the dissemination range of a single message, lacking targeted modeling for the crucial scenario of multiple messages competing for dissemination. In a competitive environment, the goal is not only to expand one's own dissemination but also to suppress or block the spread of competitors' information, requiring entirely new problem definitions and evaluation metrics. Second, in terms of technical implementation, accurately assessing the dissemination gain brought about by different combinations of content features typically requires extensive and time-consuming Monte Carlo simulations to estimate the results of the random propagation process, incurring enormous computational costs that cannot meet the urgent need for real-time response in social network information management.

[0005] In summary, existing technologies face a dual challenge in addressing information competition control under fixed propagation sources: first, existing models do not fully consider information interaction (such as priority and delay) and blocking objectives in a competitive environment; second, existing optimization algorithms have excessively high computational complexity, making them unsuitable for rapid decision-making in large-scale social networks. Therefore, there is an urgent need for a new method that can accurately characterize the dynamics of competitive propagation and efficiently compute the optimal content feature selection strategy to achieve effective, real-time blocking of competitive information propagation. Summary of the Invention

[0006] The problem addressed by this invention is how to efficiently select the content features of restrictive information under the realistic constraint of a fixed initial propagation source, so as to maximize its blocking effect on the propagation range of competing information, and overcome the problem of excessive computational complexity of traditional algorithms.

[0007] To address the aforementioned problems, this invention provides a method for blocking competitive information in social networks based on content feature selection, an electronic device, and a storage medium.

[0008] In a first aspect, the present invention provides a method for blocking competitive information in social networks based on content feature selection, comprising the following steps: S1: Construct a content-aware multi-competitive information propagation model. This model models the social network as a graph structure, associating nodes with user attributes and information with content features. The probability of information propagating along the edges is defined as being jointly determined by the basic propagation probability and the matching degree between node attributes and information content. The model supports carrying a first set of content features. Competitive information With carrying a second set of content features Restricted information Parallel propagation, and the limiting information The information is given to the competition. Higher activation priority; S2: Based on the content-aware multi-competitive information propagation model, calculate the target node in the network. The restricted information Successfully activated but not subject to the aforementioned competing information Activation blocking probability ; S3: With the goal of maximizing the total blocking probability of all nodes in the network, iteratively select from the candidate content feature set. Each content feature is added to the second content feature set. ; S4: Based on the second content feature set Generate and deploy the restriction information To suppress the competing information The spread of.

[0009] Optionally, the construction of the content-aware multi-competitive information propagation model in step S1 includes: S11: Abstract the target social network into a directed graph. ,in Represents a set of user nodes. The set of directed edges representing relationships between users; S12: Associate a set of user attributes with each user node. ; for the competition information Associated with a fixed set of first content features and for the aforementioned restriction information Associate with an initially empty second set of content features. ; S13: For either side Information from nodes propagation to nodes edge propagation probability The calculation method is as follows: ; in, For the edge The basic propagation probability, For attribute matching coefficients, This represents the number of matches between all tags of a user node and tags in the set of information content features; S14: Set the competition information With the aforementioned limiting information Starting from a fixed initial set of activated nodes and Initially, propagation proceeds according to the independent cascading rules, and the aforementioned limiting information... Compared to the aforementioned competitive information It has an r-round propagation delay; the limiting information is provided when the same node is attempted to be activated by two entities at the same time. It has priority activation rights.

[0010] Optionally, the computing node described in step S2 Blocking probability Specifically, it includes: S21: From the aforementioned competition information initial set of activated nodes Departure, for each target node Determine at least one originating from the initial set of active nodes. The path with the greatest impact is recorded, and the shortest activation distance for each path is also recorded. and path activation probability ,in ; S22: From the aforementioned restriction information initial set of activated nodes Starting from the initial set of activated nodes, construct a set containing all nodes from the initial set. To each target node And the path propagation probability is not lower than a preset threshold. Subgraph of effective propagation paths ; S23: In the subgraph In the process, the target node is calculated. At a given distance The internal restriction information Probability of activation , where the distance The value is based on the obtained shortest activation distance. ; S24: Based on the path activation probability The shortest activation distance compute nodes Regarding competitive information Source node The blocking contribution, and the integration of all nodes The contribution of the node is obtained. Total blocking probability .

[0011] Optionally, the iterative selection The content features specifically include: S31: Initialize the selected content feature set It is an empty set; S32: Execution Each iteration includes: traversing the features of currently unselected candidate content. , Temporarily added to the collection Construct a temporary set Based on the content-aware multi-competitive information propagation model and the temporary set Recalculate the total blocking probability of all nodes in the network. ; Select the sum of the blocking probabilities Candidate content features with the greatest gain It is then formally added to the selected content feature set. ; S33: The final selected content feature set is obtained. As the second set of content features .

[0012] Optionally, the user attribute set With the first content feature set Second set of content features The elements in the text are tags that represent user interests or information topics.

[0013] Optionally, calculate the target node At a given distance The internal restriction information Probability of activation Using a recursive approach: If node Belongs to the aforementioned restriction information initial set of activated nodes ,but ; If node In the subgraph If there are no incoming neighbors or D≤0, then ; Otherwise, by its subgraph All neighboring countries The activation contributions are aggregated as follows: ,in, Represents a node In subgraph The set of incoming neighbors in the middle, To restrict information From node To the node The edge propagation probability.

[0014] Optionally, the node Total blocking probability Calculated using the following formula: , in, For use in correcting nodes In competitive information From the source node The path has been reached before A factor that influences the probability of activation of other source nodes. Indicates restriction information In competitive information From node Reaching the node Required Inside the wheel, activate first The probability of.

[0015] Optionally, the propagation delay r is a non-negative integer.

[0016] Optionally, the determination of the path with the greatest influence and the construction of the subgraph in step S22 are included. The path with the maximum weight is found by using the product of the edge propagation probabilities on the path as the path weight.

[0017] In a second aspect, the present invention provides an electronic device including a processor, a communication interface, a memory, and a bus, wherein the processor, the communication interface, and the memory communicate with each other through the bus, and the processor can call logical instructions in the memory to execute the steps of the method provided in the first aspect.

[0018] Thirdly, the present invention provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the content feature-based social network competition information blocking method described in the first aspect.

[0019] The beneficial effects of the content feature-based social network competitive information blocking method of the present invention are as follows: This invention proposes a content-aware multi-competitive information propagation model that more accurately simulates the real-world situation of competing information propagation in social networks, including propagation delay and priority settings. It also supports dynamic adjustment of content attribute sets, making it suitable for scenarios requiring rapid response, such as online advertising and public opinion emergency management. The proposed blocking algorithm blocks competitive information propagation through content feature selection rather than node selection, which is more in line with practical application scenarios because social platforms can usually control the characteristics of their published content but have difficulty controlling user behavior. By designing an efficient blocking probability calculation mechanism, and through maximum impact path pruning and probability accumulation, tens of thousands of Monte Carlo simulations are avoided, significantly reducing computational complexity and achieving exponential efficiency improvements in large-scale networks while maintaining performance close to that of greedy algorithms. Attached Figure Description

[0020] Figure 1 This is a flowchart of a social network competition information blocking method based on content feature selection in an embodiment of the present invention; Figure 2 This is a flowchart illustrating the probability calculation of a node being activated by restricted information in an embodiment of the present invention. Figure 3 This is a structural block diagram of the electronic device in an embodiment of the present invention. Detailed Implementation

[0021] To better understand the purpose, technical solution, and advantages of this application, the application is described and explained below in conjunction with the accompanying drawings and embodiments.

[0022] Unless otherwise defined, the technical or scientific terms used in this application shall have the general meaning understood by one of ordinary skill in the art to which this application pertains. Words such as “a,” “an,” “an,” “the,” “the,” and “these” used in this application do not indicate quantitative limitation and may be singular or plural. The terms “comprising,” “including,” “having,” and any variations thereof used in this application are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that comprises a series of steps or modules (units) is not limited to the listed steps or modules (units) but may include steps or modules (units) not listed, or may include other steps or modules (units) inherent to these processes, methods, products, or devices. Words such as “connected,” “linked,” and “coupled” used in this application are not limited to physical or mechanical connections but may include electrical connections, whether direct or indirect. “Multiple” used in this application refers to two or more. “And / or” describes the relationship between related objects, indicating that three relationships may exist; for example, “A and / or B” can represent: A alone, A and B simultaneously, and B alone. Normally, the character " / " indicates that the objects before and after it are in an "or" relationship. The terms "first," "second," "third," etc., used in this application are merely to distinguish similar objects and do not represent a specific order of objects.

[0023] Currently, there are two main approaches to controlling information propagation in social networks. One is the traditional influence maximization method based on node selection, which focuses on choosing the optimal set of seed nodes to maximize the number of nodes ultimately activated. Some studies have extended the node selection problem to various information competition environments, but have neglected the crucial factor of users' differences in content preferences. The other approach is to maximize influence based on content feature selection. This involves starting from a fixed set of seed nodes and selecting a set of content features to construct information, thereby maximizing its expected propagation range in the network. However, current research on content features only applies to maximizing the propagation of single information and has not designed efficient algorithms for scenarios involving multiple information competitions.

[0024] 1. Existing research on maximizing influence based on content feature selection focuses on maximizing the spread of single information, lacking modeling of the key mechanism of competition among multiple information sources (especially the requirement to prioritize the activation of nodes for restricted information), resulting in insufficient scenario adaptability.

[0025] 2. For combinatorial optimization problems such as content feature selection, the most direct solution is a greedy algorithm. In content-aware models, evaluating the impact gain of each feature requires a large number of Monte Carlo simulations to estimate the propagation process, which incurs enormous computational overhead and cannot meet the real-time response requirements of social networks.

[0026] This invention provides a method for blocking competitive information in social networks based on content feature selection. It is applicable to situations where the content being spread by competitors and the initial source node are known. By optimizing the content attributes of the information being restricted, it maximizes the protection of vulnerable users in the network, thereby effectively compressing the final spread range of competitors' information.

[0027] like Figure 1 and Figure 2 As shown in the figure, a method for blocking competitive information in social networks based on content feature selection provided in this embodiment of the invention includes the following steps: S1: Construct a content-aware multi-competitive information propagation model. The model models the social network as a graph structure, associating nodes with user attributes and information with content features. The probability of information propagating along the edges is defined as being jointly determined by the basic propagation probability and the matching degree between node attributes and information content. The model supports carrying a first set of content features. Competitive information With carrying a second set of content features Restricted information Parallel propagation, and information restriction Given more competitive information Higher activation priority; Specifically, step S1 involves constructing a content-aware multi-competitive information propagation model, which includes the following steps: S11: Abstract the target social network into a directed graph. ,in Represents a set of user nodes. The set of directed edges representing relationships between users; S12: For each user node Associate a set of user attributes For competitive information Associated with a fixed set of first content features and to restrict information Associate with an initially empty second set of content features. It should be noted that this set of user attributes Composed of several interest / topic tags, the first set of content features Second content feature set Its elements are also topic tags. First set of content features It contains a set of fixed content attributes, while the second set of content features... It is initially an empty set.

[0028] User attribute set With the first set of content features Second set of content features The elements in the text are tags that represent user interests or information topics.

[0029] S13: For either side Information from nodes propagation to nodes edge propagation probability Defined as: ;in .

[0030] in, For the edge The basic propagation probability, For attribute matching coefficients, This represents the number of matches between all tags of a user node and tags in the set of information content features; this formula ensures that the higher the tag matching degree, the greater the probability of propagation.

[0031] S14: Regarding the propagation rules, a variant of the independent cascade model is adopted, where each node, after being activated, has one chance to... Activate its neighbors; set competition information With Restricted Information Each starts from its own fixed initial set of activated nodes. and Initially, propagation follows independent cascading rules, with restrictions set. Compared to competitive information It has a propagation delay of r rounds; it should be noted that the propagation delay r is a non-negative integer. When the same node is attempted to be activated by two entities at the same time, the information is limited. It has priority activation rights to reflect its protective function.

[0032] S2: A content-aware multi-competitive information propagation model for calculating target nodes in the network. Restricted Information Successfully activated and not contested. Activation blocking probability ; Calculate the blocking probability of the target node; define the node. Blocking probability For its restricted information Successfully activated and not contested. The probability of activation; for efficient calculation, this invention adopts an approximation mechanism based on the maximum influence path pruning; specifically, S2 calculates the blocking probability of the target node by the following steps: S21: From competitive information All initial active node sets Starting point, for each target node in the network. Determine at least one originating from the initial set of active nodes. The path with the greatest impact is recorded, and the shortest activation distance for each path is also recorded. and path activation probability ,in ; Specifically, from competitive information All initial active node sets Starting point, for each target node in the network. Construct the set of paths with the greatest impact. Specifically, for each source node... Calculate from arrive The largest impact path MIPC The overall propagation probability for each path is not lower than a preset threshold. Record the shortest activation distance of the MIPC path. and activation probability .

[0033] S22: From restrictive information initial set of activated nodes Starting from the initial set of activated nodes, construct a set containing all nodes from the initial set. To each target node And the path propagation probability is not lower than a preset threshold. Subgraph of effective propagation paths ; Determine the path of maximum influence and construct the subgraph in step S22 Both methods employ a variant of Dijkstra's algorithm, using the product of the edge propagation probabilities along the path as the path weight to find the path with the maximum weight.

[0034] Specifically, regarding information restrictions Similarly, starting from its initial source node set, construct a pruned subgraph. This subgraph contains all nodes from the source node to the target node. And the path probability is not lower than a preset threshold. Effective dissemination paths; from restricting information initial set of activated nodes Starting from this point, construct a pruned subgraph. This subgraph Includes all nodes from the initial active node set From any source node in the network to every target node in the network Furthermore, the overall propagation probability of the MIPL, the path with the greatest impact, is not lower than the same threshold. Effective propagation paths; subgraph The construction is also based on the maximum impact path mechanism, ensuring that only high-probability propagation paths are retained to improve computational efficiency.

[0035] S23: In the subgraph In the process, the target node is calculated. At a given distance Internal restricted information Probability of activation , where the distance The value is based on the obtained shortest activation distance. ; Specifically, in the subgraph In the process, each node is calculated using a bottom-up recursive approach. Within a distance of Within the range The probability of activation is denoted as Among them, distance The value comes from the shortest activation distance calculated in step S21. .

[0036] S24: Based on path activation probability Shortest activation distance compute nodes Regarding competitive information Source node The blocking contribution, and the integration of all nodes The contribution of the node is obtained. Total blocking probability .

[0037] Calculate the target node At a given distance Internal restricted information Probability of activation Using a recursive approach: If node This is restricted information. initial set of activated nodes ,but ; If node In subgraph If there are no incoming neighbors or D≤0, then ; Otherwise, by its subgraph All neighboring countries The activation contributions are aggregated as follows: ,in, Represents a node In subgraph The set of incoming neighbors in the middle, To restrict information From node To the node The edge propagation probability.

[0038] Specifically, the target node At a given distance Internal restricted information Probability of activation The calculation is as follows Figure 2 As shown, the following recursive rule applies: if the node It is itself a restriction on information The initial activation node (i.e. ),but If node In subgraph There are no incoming neighbors, or the currently calculated propagation distance has reached or exceeded the given distance. ,but ;otherwise, Because of its subgraph All neighboring countries The activation contributions are aggregated as follows: ,in: Represents a node In subgraph The set of incoming neighbors in the middle, It is a restriction of information From node propagation to nodes The edge propagation probability.

[0039] Combining the results of steps S21 and S23, node The actual blocking probability is calculated as follows: ,in: Approximately represents competitive information From the source node First successful node activation The probability, It is a correction factor used to account for the period before path MIPC(u, v) is activated. Already from The probability of other source nodes being activated, and This represents restrictive information. In competitive information From node Reaching the node Required Within the round, activate the node first. The probability; the core idea of ​​this formula is that only when information is restricted... In competitive information Preemptively activate nodes within the activation path length Only by traversing all possible competing information can effective blocking be achieved. source node By accumulating their contributions, the node can be obtained. Overall blocking probability .

[0040] S3: With the goal of maximizing the total blocking probability of all nodes in the network, iteratively select from the candidate content feature set. One content feature is added to the second content feature set. ; Iterative selection The content features specifically include: S31: Initialize the selected content feature set It is an empty set; S32: Execution Each iteration includes: traversing the features of currently unselected candidate content. Content characteristics Temporarily added to the collection Construct a temporary set Content-aware multi-competitive information propagation model and temporary set Recalculate the total blocking probability of all nodes in the network. Choose the option that sums the blocking probabilities. Candidate content features with the greatest gain It was formally added to the set of selected content features. ; S33: The final set of selected content features As a second set of content features .

[0041] Specifically, the optimal content features are selected iteratively; and selected from all candidate topic tags. The optimal tags are added to the second content feature set. To maximize the overall network blocking effect, the S3 iteration selects the optimal content features through the following steps: Initialize the selected tag set .

[0042] In each iteration, iterate through all unselected tags. Temporarily join in get .

[0043] based on Recalculate the total probability of network-wide blocking. .

[0044] Choose to The label with the highest gain to formally join .

[0045] Repeat the above process until... .

[0046] S4: Based on the second content feature set Generate and deploy restriction information To suppress competing information The spread of.

[0047] Specifically, deploy and initiate restrictions on information dissemination; select the steps in step S3. Each label is configured to restrict information. Second content feature set and from the preset initial propagation source Initiate its spread. Due to Optimized, information restricted Will try to be in Activating susceptible nodes before they arrive effectively blocks their propagation path, thus restricting competing information.

[0048] In summary, the content-aware multi-competitive information propagation model proposed in this invention more accurately simulates the real-world situation of competitive information propagation in social networks, including propagation delay and priority settings. It also supports dynamic adjustment of content attribute sets, making it suitable for scenarios requiring rapid response, such as online advertising and public opinion emergency management. The blocking algorithm proposed in this invention blocks competitive information propagation through content feature selection rather than node selection, which is more in line with practical application scenarios because social platforms can usually control the characteristics of their published content but have difficulty controlling user behavior. This invention designs an efficient blocking probability calculation mechanism, avoiding tens of thousands of Monte Carlo simulations through maximum impact path pruning and probability accumulation, significantly reducing computational complexity and achieving exponential efficiency improvements in large-scale networks while maintaining performance close to that of greedy algorithms.

[0049] like Figure 3 As shown in the figure, an electronic device provided by an embodiment of the present invention includes: a processor 610, a communication interface 620, a memory 630, and a communication bus 640, wherein the processor 610, the communication interface 620, and the memory 630 communicate with each other through the communication bus 640. The processor 610 can call logical instructions in the memory 630 to execute the following method: S1: Construct a content-aware multi-competitive information propagation model. This model models the social network as a graph structure, associating nodes with user attributes and information with content features. The probability of information propagating along the edges is defined as being jointly determined by the basic propagation probability and the matching degree between node attributes and information content. The model supports carrying a first set of content features. Competitive information With carrying a second set of content features Restricted information Parallel propagation, and the limiting information The information is given to the competition. Higher activation priority; S2: Based on the content-aware multi-competitive information propagation model, calculate the target node in the network. The restricted information Successfully activated but not subject to the aforementioned competing information Activation blocking probability ; S3: With the goal of maximizing the total blocking probability of all nodes in the network, iteratively select from the candidate content feature set. Each content feature is added to the second content feature set. ; S4: Based on the second content feature set Generate and deploy the restriction information To suppress the competing information The spread of.

[0050] Furthermore, the logical instructions in the aforementioned memory 630 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0051] This invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, is implemented to perform the methods provided in the above embodiments.

[0052] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0053] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for blocking competitive information in social networks based on content feature selection, characterized in that, Includes the following steps: S1: Construct a content-aware multi-competitive information propagation model. This model models the social network as a graph structure, associating nodes with user attributes and information with content features. The probability of information propagating along the edges is defined as being jointly determined by the basic propagation probability and the matching degree between node attributes and information content. The model supports carrying a first set of content features. Competitive information With carrying a second set of content features Restriction information Parallel propagation, and the limiting information The information given is more competitive than the aforementioned information. Higher activation priority; S2: Based on the content-aware multi-competitive information propagation model, calculate the target node in the network. The restricted information Successfully activated but not subject to the aforementioned competing information Activation blocking probability ; S3: With the goal of maximizing the total blocking probability of all nodes in the network, iteratively select from the candidate content feature set. Each content feature is added to the second content feature set. ; S4: Based on the second content feature set Generate and deploy the restriction information To suppress the competing information The spread of.

2. The social network competitive information blocking method based on content feature selection according to claim 1, characterized in that, The construction of the content-aware multi-competitive information propagation model in step S1 includes: S11: Abstract the target social network into a directed graph. ,in Represents a set of user nodes. A set of directed edges representing relationships between users; S12: Associate a set of user attributes with each user node. ; for the competition information Associated with a fixed set of first content features and for the aforementioned restriction information Associate with an initially empty second set of content features. ; S13: For either side Information from nodes propagation to nodes edge propagation probability The calculation method is as follows: ; in, For the edge The basic propagation probability, For attribute matching coefficients, This represents the number of matches between all tags of a user node and tags in the set of information content features; S14: Set the competition information With the aforementioned limiting information Starting from a fixed initial set of activated nodes and Initially, propagation proceeds according to the independent cascading rules, and the aforementioned limiting information... Compared to the aforementioned competitive information It has an r-round propagation delay; the limiting information is provided when the same node is attempted to be activated by two entities at the same time. It has priority activation rights.

3. The social network competitive information blocking method based on content feature selection according to claim 1, characterized in that, The computing node in step S2 Blocking probability Specifically, it includes: S21: From the aforementioned competition information initial set of activated nodes Departure, for each target node Determine at least one originating from the initial set of activated nodes. The path with the greatest impact is recorded, and the shortest activation distance for each path is also recorded. and path activation probability ,in ; S22: From the aforementioned restriction information initial set of activated nodes Starting from the initial set of activated nodes, construct a set containing all nodes from the initial set. To each target node And the path propagation probability is not lower than a preset threshold. Subgraph of effective propagation paths ; S23: In the subgraph In the process, the target node is calculated. At a given distance The internal restriction information Probability of activation , where the distance The value is based on the obtained shortest activation distance. ; S24: Based on the path activation probability The shortest activation distance compute nodes Regarding competitive information Source node The blocking contribution, and the integration of all nodes The contribution of the node is obtained. Total blocking probability .

4. The social network competitive information blocking method based on content feature selection according to claim 1, characterized in that, The iterative selection The content features specifically include: S31: Initialize the selected content feature set It is an empty set; S32: Execution Each iteration includes: traversing the features of currently unselected candidate content. , Temporarily added to the collection Construct a temporary set Based on the content-aware multi-competitive information propagation model and the temporary set Recalculate the total blocking probability of all nodes in the network. ; Select the sum of the blocking probabilities Candidate content features with the greatest gain It is then formally added to the selected content feature set. ; S33: The final selected content feature set is obtained. As the second set of content features .

5. The social network competitive information blocking method based on content feature selection according to claim 1, characterized in that, The user attribute set With the first content feature set Second set of content features The elements in the text are tags that represent user interests or information topics.

6. The social network competitive information blocking method based on content feature selection according to claim 3, characterized in that, Calculate the target node At a given distance The internal restriction information Probability of activation Using a recursive approach: If node Belongs to the aforementioned restriction information initial set of activated nodes ,but ; If node In the subgraph If there are no incoming neighbors or D≤0, then ; Otherwise, by its subgraph All neighboring countries The activation contributions are aggregated as follows: ,in, Represents a node In subgraph The set of incoming neighbors in the middle, To restrict information From node To the node The edge propagation probability.

7. The social network competitive information blocking method based on content feature selection according to claim 3, characterized in that, The node Total blocking probability Calculated using the following formula: ， in, For use in correcting nodes In competitive information From the source node The path was reached before A factor that determines the probability of activation of other source nodes. Indicates restriction information In competitive information From node Reaching the node Required Inside the wheel, activate first The probability of.

8. The social network competitive information blocking method based on content feature selection according to claim 2, characterized in that, The propagation delay r is a non-negative integer.

9. The social network competitive information blocking method based on content feature selection according to claim 2, characterized in that, The determination of the path with the greatest influence and the construction of the subgraph in step S22 The path with the maximum weight is found by using the product of the edge propagation probabilities on the path as the path weight.

10. An electronic device, characterized in that, Including memory and processor; The memory is used to store computer programs; The processor is configured to, when executing the computer program, implement the social network competition information blocking method based on content feature selection as described in any one of claims 1 to 9.

Images